Markov chain in SAS for clickstream

vrushankshah — Sun, 22 Jul 2018 17:20:55 GMT

I have a dataset which contains various clicks which a user did while going through the website. It also contains uniqueid of each user, date and time of the click and their name. Essentially my main aim is to predict what will be next click of the user. For example, if it starts from home then goes to clothing then goes to Denim then from Denim what is the highest probability of its next click?There are 50000 unique click patterns in a month which a user has clicked. Will Markov chain be feasible for such data?

Clicks Time Id
Home,Clothing,Men,Denim,Home 02/10/2018/3:22pm 1234
Home,Kitchenware,glass,purchase 03/10/2018/4:00pm 4567
Home,Clothing,Men,Denim,Purchase 04/10/2018/3:55pm 7891
Home,Clothing,Men 05/10/2018/2:56pm 6789

Re: Markov chain in SAS for clickstream

Ksharp — Mon, 23 Jul 2018 12:48:55 GMT

It is more like Market Basket Analysis.Change your data like:

Home,Clothing,Men,Denim,Home 1234

-->

Home,Clothing, 1234

Clothing ,Men, 1234

Men,Denim, 1234

Denim,Home 1234

After that count the frequency.

If MC could apply to it , calling @Rick_SAS

topic Re: Markov chain in SAS for clickstream in SAS Procedures

Markov chain in SAS for clickstream

Re: Markov chain in SAS for clickstream